The term Hybrid as such does not convey too much. It gains significance when the physical storage methods become flexible, meaning: They depend on the category of data, configurable or at choice of the client on dynamic request:

• Volatile data with high update frequency goes into memory, because even the fastest SSD cannot support millions of of updates per second.
• Relatively static with medium update frequency is stored to an SSD. Still it is cached in RAM.
• Mass data with more or less read only access (video, audio content delivery) is stored on cheap magnetic drives. Also it is cached in RAM.
• Configurable per object class.
• BLOB support configurable per attribute type. Storage on Disk only. Dynamically fetched into main memory only on demand and only for the duration of the request.

Our database supports storage configuration by means of schema directives as well as dynamic assignment on behalf of client preferences, all dependent on the category of data. The whole storage sub system is optimized to make use of as many parallel devices as present in the machine. Space can be extended via NAS systems, there is no limit in static or dynamic capacity. The physical storage layer is data model agnostic. Clients see the same data model, regardless of the chosen storage method. Alltogether this creates an unmatched combination of capacity and performance.